fix import

utils/stt.py

@@ -4,6 +4,7 @@ Handles audio preprocessing and transcription
 """
 
 import logging
+import numpy as np
 logger = logging.getLogger(__name__)
 
 import torch